C

libcurl与CLOSE_WAIT

2010-03-02 13:41 | Posted by Sunshow | Filed Under C, Network

调用libcurl下载，然后使用netstat查看发现有大量的TCP连接保持在CLOSE_WAIT状态
查看libcurl的文档说明，有这样一个选项：

CURLOPT_FORBID_REUSE

Pass a long. Set to 1 to make the next transfer explicitly close the connection when done. Normally, libcurl keeps all connections alive when done with one transfer in case a succeeding one follows that can re-use them. This option should be used with caution and only if you understand what it does. Set to 0 to have libcurl keep the connection open for possible later re-use (default behavior).

也就是说，默认情况下libcurl完成一个任务以后，出于重用连接的考虑不会马上关闭
如果没有新的TCP请求来重用这个连接，那么只能等到CLOSE_WAIT超时，这个时间默认在7200秒甚至更高，太多的CLOSE_WAIT连接会导致性能问题

解决方法：

curl_easy_setopt(curl, CURLOPT_FORBID_REUSE, 1);

最好再修改一下TCP参数调低CLOSE_WAIT和TIME_WAIT的超时时间

标签：libcurl, TCP, 下载

url的encode和decode

2009-11-14 16:34 | Posted by Sunshow | Filed Under C

/* ---------------------------------------------------------------------------
* Encode URL by converting special characters to %XX (where XX are hexadecimal digits)
* Don't forget to free the return value.
*/
char *urlencode(const char *url)
{
#define COPY_TO_ENCODE_URL(c) \
if (outlen < pos) { \
outlen += 10; \
out = realloc(out, outlen + 1); \
} \
out[pos ++] = c; \
size_t i, len = strlen(url), outlen;
outlen = len;
char *out = malloc(outlen + 1);
int pos;
for (i = 0, pos = 0; i < len; i ++) {
if ((url[i] >= 48 && url[i] <= 57) // 0-9
|| (url[i] >= 65 && url[i] <= 90) //a-z
|| (url[i] >= 97 && url[i] <= 122) //A-Z
|| url[i] == ';' || url[i] == '/' || url[i] == '?'
|| url[i] == ':' || url[i] == '@' || url[i] == '&'
|| url[i] == '=' || url[i] == '+' || url[i] == '$'
|| url[i] == ',' || url[i] == '-' || url[i] == '_'
|| url[i] == '.' || url[i] == '!' || url[i] == '~'
|| url[i] == '*' || url[i] == '\'' || url[i] == '('
|| url[i] == ')') {
/* straight copy */
COPY_TO_ENCODE_URL(url[i]);
}
else {
#undef HEX_TO_DIGIT
char dig1 = (url[i] & 0xF0) >> 4;
char dig2 = (url[i] & 0x0F);
if (dig1 >= 0 && dig1 <= 9) dig1 += 48; //0,48inascii
if (dig1 >= 10 && dig1 <= 15) dig1 += 65 - 10; //A,65inascii
if (dig2 >= 0 && dig2 <= 9) dig2 += 48;
if (dig2 >= 10 && dig2 <= 15) dig2 += 65 - 10;
COPY_TO_ENCODE_URL('%');
COPY_TO_ENCODE_URL(dig1);
COPY_TO_ENCODE_URL(dig2);
}
}
out[pos] = '\0';
return (out);
}
/* ---------------------------------------------------------------------------
* Decode URL by converting %XX (where XX are hexadecimal digits) to the
* character it represents. Don't forget to free the return value.
*/
char *urldecode(const char *url)
{
size_t i, len = strlen(url);
char *out = malloc(len + 1);
int pos;
for (i = 0, pos = 0; i < len; i ++) {
if (url[i] == '%' && i+2 < len &&
isxdigit(url[i + 1]) && isxdigit(url[i + 2])) {
/* decode %XX */
#define HEX_TO_DIGIT(hex) ( \
((hex) >= 'A' && (hex) <= 'F') ? ((hex) - 'A' + 10): \
((hex) >= 'a' && (hex) <= 'f') ? ((hex) - 'a' + 10): \
((hex) - '0') )
out[pos ++] = HEX_TO_DIGIT(url[i + 1]) * 16 +
HEX_TO_DIGIT(url[i + 2]);
i += 2;
#undef HEX_TO_DIGIT
}
else {
/* straight copy */
out[pos ++] = url[i];
}
}
out[pos] = '\0';
return (out);
}

标签：decode, encode, url

[Socket]获取客户端IP

2009-07-16 13:41 | Posted by Sunshow | Filed Under C

有两种方法，一种是在accept的时候获取，一种是通过getpeername获取

#include <sys/socket.h>

int accept(int socket, struct sockaddr *restrict address,
socklen_t *restrict address_len);

int getpeername(int socket, struct sockaddr *restrict address,
socklen_t *restrict address_len);

两者区别主要在取IP的fd不一样，前者是在监听的fd，后者是在连接建立的fd

struct sockaddr_in addr;
socklen_t addr_len;
int32_t listen_fd, sock_fd;

/* create listening port */

addr_len = sizeof(addr);

memset(&addr, 0, addr_len);
sock_fd = accept(listen_fd, (struct sockaddr *)&addr, &addr_len);
printf("%d\n", addr.sin_addr.s_addr);

memset(&addr, 0, addr_len);
getpeername(sock_fd, (struct sockaddr *)&addr, &addr_len);
printf("%d\n", addr.sin_addr.s_addr);

当然这种长整型格式的IP不一定是我们想要的，可以通过inet_ntoa转换